pdf-to-markdown
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of entire PDF documents into the agent's active context. This creates a surface for indirect prompt injection where a malicious document could contain instructions intended to override the agent's behavior.\n
- Ingestion points: The
scripts/pdf_to_md.pytool reads arbitrary user-provided PDF files.\n - Boundary markers: While the output includes a YAML metadata header, there are no explicit delimiters or instructions for the agent to treat the document body as untrusted data.\n
- Capability inventory: The skill allows for file reading, file writing (converted markdown and images), and directory deletion (cache management).\n
- Sanitization: The skill converts PDF structure to Markdown but does not perform sanitization of the underlying text content for injection patterns.\n- [EXTERNAL_DOWNLOADS]: The IBM Docling library used by this skill downloads AI models from Hugging Face during its initial execution. Hugging Face is a well-known and trusted service in the AI community.
Audit Metadata